Research On Video Coding Techniques For Mobile Devices

Video coding is one of the key techniques of multimedia communication. With the development of computer and Internet technologies, more and more video-based applications come into being and various video-based systems become maturity. At the same time, video coding techniques not only focus on high coding efficiency, but also consider high quality and high stability.Recently, with the rapid development of wireless networks and consumer electronics, users are expecting to obtain multimedia information at anytime and anywhere. The latest wireless communication techniques, such as WLAN, Bluetooth and 3G, can provide basic bandwidth for multimedia communication, as well as real-time communication of compressed video. And the computing power of embedded system can satisfy the basic requirement of real-time multimedia signal processing.However, due to the constraint and diversity of computing resources on mobile devices as well as dithered bandwidth and instability of wireless networks, traditional video coding techniques can not be directly used in mobile environment. There are many technical problems and challenges to achieve real-time video applications on mobile devices. Therefore, it is important to research the techniques for video coding in mobile environment. The thesis pays more attention on these key techniques and the main research work is as follows:First, combining with the basic techniques of video coding and development of video coding standards, we analyze the current research topics and key problems of video coding, and then present our research issues.Second, we analyze the key algorithms of the latest video coding standard H.264, also with the complexity. Although the high coding efficiency of H.264 can satisfy the requirement of wireless bandwidth, the high complex algorithms can not adapt to computing power of mobile devices. Thus, we design and optimize the highest complex algorithms of H.264, propose the fast variable-block-size motion estimation algorithm and predictive mode decision algorithm. These algorithms can effectively reduce the complexity of video encoding for H.264.Third, considering the diversity of computing resources of mobile devices, we propose the complexity-scalable video coding scheme. We mainly discuss the complexity-scalability for motion estimation, which is the most complex module of video coding, and then propose the complexity-configurable and complexity-bounded motion estimation algorithms, which can satisfy real-time video coding for mobile devices with various computing resources. Based on the research, we present a framework for fine-granular complexity-scalable motion estimation, and achieve the tradeoff between the accuracy of motion estimation and its computational complexity.Next, we research the error resilient techniques of video coding in fluctuant instable wireless environment, and propose two efficient error resilient algorithms. The one is based on significant block protection. It adopts unbalanced error-resilience which can effectively protect the region of interest (ROI), thus improves the subjective video quality efficiently. The other is adaptive error recovery algorithm, which is based on multiple description video coding. The algorithm not only is compatible with existing video coding standard, but also has low-complexity and good error-resilient features.Finally, according to the features of mobile devices and embedded systems, we propose the component-based video coding framework, which can satisfy the requirement of mobile video coding applications. Based on the framework and combining with our aforementioned research, we realize the practical low complexity real-time video codec, which can be used for real-time video coding applications on existing mobile devices.